TL;DR:
- Vast Data, a pioneer in flash storage and NFS protocol optimization, is poised for a blockbuster IPO in 2024 or 2025, with a valuation of $9.1 billion.
- Their Vast Data Platform addresses the need for high-performance storage in HPC, simulation, modeling, and AI training.
- The recent Series E funding round raised $118 million, bringing total funding to $381 million.
- Vast Data has tripled its business volume in the past year, achieving positive cash flow and 90% gross margins.
- Key customers include Lawrence Livermore National Laboratory, Texas Advanced Computing Center, and AI cloud builders like CoreWeave and Lambda.
- Vast Data’s strategic focus on large-scale customers and substantial deals has been a winning formula.
- Vast Data’s mission is to democratize AI technology, making it accessible to a broader range of organizations.
- The company’s vision includes the potential introduction of Vast Compute and Vast Network to complement its storage offerings.
Main AI News:
In the realm of high-tech enterprises and their quest for funding, the alphabetic series often stops short at Series E. If, by that stage, a company hasn’t yet defined its identity or if the market hasn’t bestowed recognition, it typically falls into the 80 percent of enterprises that don’t survive, rather than joining the exclusive 20 percent that either secures an acquisition deal with a larger corporation, venture capitalist backing, or embark on a groundbreaking initial public offering on Wall Street.
If we were to place a speculative wager, which we can’t, as we don’t engage in financial investments, we’d bet heavily on Vast Data. This trailblazing company, renowned for its prowess in optimizing flash storage and the NFS protocol while seamlessly bridging the realms of storage and databases, seems poised not only to survive but to become one of the blockbuster IPOs of either 2024 or, possibly, 2025, contingent upon the global economic landscape.
In today’s landscape of high-performance, scale-out workloads, a distinctive kind of storage solution is requisite – one that harmonizes effectively with HPC simulations, modeling, and AI training. Enter the Vast Data Platform, the latest evolution in storage solutions offered by Vast Data. With the recent infusion of $118 million in its Series E funding round, Vast Data’s total funding vaults to an impressive $381 million, while its valuation soars to an impressive $9.1 billion, according to co-founder Jeff Denworth. Over the past two years, this valuation has surged 2.5 times, solidifying Vast Data’s standing as the preferred storage platform for large-scale infrastructures.
Lawrence Livermore National Laboratory stands as a prominent early adopter of Vast Data’s offerings. This comes as no surprise, considering that some of the company’s founders played pivotal roles in the creation of the Lustre parallel file system, which gained prominence at this flagship U.S. Department of Energy HPC center. Subsequently, the Texas Advanced Computing Center has deployed Vast Data for its “Stampede 3” supercomputer’s scratch file system, while AI cloud pioneers like CoreWeave, G42, and Lambda have harnessed Vast Data’s back-end storage prowess to power their GPU clusters.
This strategic approach aligns with the principles articulated by Denworth and co-founder Renen Hallack back in February 2019 when Vast Data emerged from stealth mode. In stark contrast to Nutanix, Pure Storage, and previous storage startups, Vast Data has maintained a steadfast focus on large-scale customers and the pursuit of substantial deals. This strategy has proven to be the swiftest path to product enhancement and profitability. We eagerly anticipate examining Vast Data’s S-1 filing when it takes the plunge into the public market, as it will provide insight into whether this approach has borne fruit.
“We have several dozen cloud providers as customers, but that doesn’t encapsulate our entire business,” explains Denworth, who, after donning multiple hats, now bears the title of co-founder. “Many enterprises are purchasing substantial AI systems, but AI merely represents the spearhead of our comprehensive portfolio.”
Denworth is discreet about disclosing specific capacity figures accumulated over the past four years, but he estimates the range to fall between 10 exabytes and 20 exabytes. He adds that the past year has witnessed a tripling of business volume and a corresponding valuation surge. Over the past two years, Vast Data has transitioned into a software-defined storage vendor after initially peddling complete hardware and software appliances for its first two years of existence. Cumulatively, Vast Data has booked more than $1 billion in Vast Data Platform software sales, and over the past twelve months, it has expanded the company’s size by 3.3 times while maintaining a positive cash flow and boasting gross margins approaching 90 percent.
Evidently, the “go big and go strong” strategy, devised by Denworth and Hallack from the outset, has yielded remarkable results.
Denworth identifies only five AI companies, valued at a minimum of $5 billion, that have witnessed their valuations more than double in 2023. This elite group comprises Anthropic, OpenAI, Nvidia, CoreWeave, and Vast Data. Remarkably, Vast Data experienced an upround during its recent funding round, bucking the trend observed among Silicon Valley unicorns.
The funding round appears to be less about capital injection and more about elevating the company’s valuation, a strategic move that often precedes a high-profile IPO. Fidelity Management & Research Company, the investment arm of Fidelity Investments, has joined the board – a customary move in anticipation of hotly anticipated IPOs. Notably, New Enterprise Associates, BOND Capital, and Drive Capital have also contributed to this funding round. As a fascinating aside, Denworth discloses that the Series B, Series C, and Series D funds raised by Vast Data in prior years are currently accruing interest in the company’s coffers. The prospect of Vast Data’s S-1 filing, therefore, looms enticingly on the horizon.
While consensus prevails within the AI compute realm, the storage domain remains marked by a lack of unanimity.
“Nvidia envisions a world where bespoke storage solutions are procured for training and inference purposes,” elucidates Denworth. “Our vision, in contrast, revolves around enabling customers to seamlessly integrate AI into their existing data infrastructure, eliminating the need for specialized AI-centric storage. Imagine different applications operating seamlessly on a unified network, with data accessibility optimized throughout. We’re currently servicing clients with hundreds of petabytes of capacity, exploring the application of large language models across colossal datasets. Copying vast datasets to dedicated local storage is an inefficient proposition.”
It’s plausible that Vast Data may inevitably need to introduce a Vast Compute counterpart, interconnected via a Vast Network. An IPO could very well furnish Vast Data with the resources needed to expand its mission and potentially lead to a transformation, culminating in a rebranded identity as Vast Systems. Hallack’s vision, articulated back in August when Vast Data Platform’s database extensions were unveiled, holds enduring relevance:
“Eight years ago, AI was relegated to identifying cats in YouTube videos – a far cry from its present-day significance. We foresaw that AI would play a pivotal role in the IT sector’s evolution over the next two decades and sought to democratize this transformative technology, making it accessible to a broader spectrum of organizations. Our mission remains unwavering – to ensure that AI is ubiquitous, rather than the exclusive domain of a select few.”
Conclusion:
Vast Data’s rapid growth, substantial funding, and strategic focus on large-scale customers position it as a prominent player in the AI-driven storage market. Its commitment to democratizing AI and potential expansion into compute and networking signify a promising future for the company, reflecting the increasing demand for integrated AI solutions in the business landscape.
